Few papers discuss how the economic burden of patients with stroke receiving rehabilitation courses is related to post-acute care (PAC) programs. This is the first study to explore the economic burden of stroke patients receiving PAC rehabilitation and to evaluate the impact of multidisciplinary PAC programs on cost and functional status simultaneously. A total of 910 patients with stroke between March 2014 and October 2018 were separated into a PAC group (at two medical centers) and a non-PAC group (at three regional hospitals and one district hospital) by using propensity score matching (1:1). A cost–illness approach was employed to identify the cost categories for analysis in this study according to various perspectives. Total direct medical cost in the per-diem-based PAC cohort was statistically lower than that in the fee-for-service-based non-PAC cohort (p < 0.001) and annual per-patient economic burden of stroke patients receiving PAC rehabilitation is approximately US $354.3 million (in 2019, NT $30.5 = US $1). Additionally, the PAC cohort had statistical improvement in functional status vis-à-vis the non-PAC cohort and total score of each functional status before rehabilitation and was also statistically significant with its total score after one-year rehabilitation training (p < 0.001). Early stroke rehabilitation is important for restoring health, confidence, and safe-care abilities in these patients. Compared to the current stroke rehabilitation system, PAC rehabilitation shortened the waiting time for transfer to the rehabilitation ward and it was indicated as an efficient policy for treatment of stroke in saving medical cost and improving functional status.

Few studies have investigated the characteristics of stroke inpatients after post-acute care (PAC) rehabilitation, and few studies have applied propensity score matching (PSM) in a natural experimental design to examine the longitudinal impacts of a medical referral system on functional status. This study coupled a natural experimental design with PSM to assess the impact of a medical referral system in stroke patients and to examine the longitudinal effects of the system on functional status. The intervention was a hospital-based, function oriented, 12-week to 1-year rehabilitative PAC intervention for patients with cerebrovascular diseases. The average duration of PAC in the intra-hospital transfer group (31.52 days) was significantly shorter than that in the inter-hospital transfer group (37.1 days) (p < 0.001). The intra-hospital transfer group also had better functional outcomes. The training effect was larger in patients with moderate disability (Modified Rankin Scale, MRS = 3) and moderately severe disability (MRS = 4) compared to patients with slight disability (MRS = 2). Intensive post-stroke rehabilitative care delivered by per-diem payment is effective in terms of improving functional status. To construct a vertically integrated medical system, strengthening the qualified local hospitals with PAC wards, accelerating the inter-hospital transfer, and offering sufficient intensive rehabilitative PAC days are the most essential requirements.

In this large-scale prospective cohort study, a propensity score matching method was applied in a natural experimental design to investigate how post-acute care (PAC) after stroke affects functional status and to identify predictors of functional status. The main objective of this study was to examine longitudinal changes in various measures of functional status in stroke patients and predictors of scores for these measures before and after PAC. A group of patients who had received PAC for stroke at one of two medical centers (PAC group, n = 273) was compared with a group who had received standard care for stroke at one of four hospitals (three regional hospital and one district hospital; non-PAC group, n = 273) in Taiwan from March, 2014, to October, 2018. The patients completed the functional status measures before rehabilitation, the 12th week and the 1st year after rehabilitation. Generalized estimating equations were used to estimate differences-in-differences models for examining the effects of PAC. The average age was 68.0 (SD = 8.1) years, and males accounted for 57.9%. During the follow-up period, significant risk factors for poor functional outcomes were advanced age, hemorrhagic stroke, and poor function scores before rehabilitation (p < 0.05). Between-group comparisons at subsequent time points revealed significantly higher functional status scores in the PAC group versus the non-PAC group (p < 0.001). Notably, for all functional status measures, between-group differences in total scores significantly increased over time from baseline to 1 year post-rehabilitation (p < 0.001). The contribution of this study is its further elucidation of the clinical implications and health policy implications of rehabilitative care after stroke. Specifically, it improves understanding of the effects of PAC in stroke patients at different follow-up times. Therefore, a policy implication of this study is that standard care for stroke should include intensive rehabilitative PAC to maximize recovery of overall function.

Background No studies have discussed machine learning algorithms to predict the risk of 30-day readmission in patients with stroke. The objective of the present study was to compare the accuracy of the artificial neural network (ANN), K nearest neighbor (KNN), support vector machine (SVM), naive Bayes classifier (NBC), and Cox regression (COX) models and to explore the significant factors in predicting 30-day readmission after stroke. Methods This study prospectively compared the accuracy of the models using clinical data for 1,476 patients with stroke treated in six hospitals between March, 2014 and September, 2019. A training dataset (n=1,033) was used for model development, a testing dataset (n=443) was used for internal validation, and a validating dataset (n=167) was used for external validation. A global sensitivity analysis was performed to compare the significance of the selected input variables. Results Of all forecasting models, the ANN model had the highest accuracy in predicting 30-day readmission after stroke and had the highest overall performance indices. According to the ANN model, 30-day readmission was significantly associated with post-acute care (PAC) program, patient attributes, clinical attributes, and functional status scores before re-habilitation (all P <0.05). Additionally, PAC program was the most significant variable affecting 30-day readmission, followed by nasogastric tube insertion, and stroke type ( P <0.05). Conclusions Comparisons of the five forecasting models indicated that the ANN model had the highest accuracy in predicting 30-day readmission in stroke patients. Before stroke patients are discharged from hospitalization, they should be counseled regarding their potential for recovery and other possible outcomes. These important predictors can also be used to educate candidates for stroke patients who underwent PAC rehabilitation with respect to the course of recovery and health outcomes.

BackgroundMachine learning algorithms for predicting 30-day stroke readmission are rarely discussed. The aims of this study were to identify significant predictors of 30-day readmission after stroke and to compare prediction accuracy and area under the receiver operating characteristic (AUROC) curve in five models: artificial neural network (ANN), K nearest neighbor (KNN), random forest (RF), support vector machine (SVM), naive Bayes classifier (NBC), and Cox regression (COX) models.MethodsThe subjects of this prospective cohort study were 1,476 patients with a history of admission for stroke to one of six hospitals between March, 2014, and September, 2019. A training dataset (n = 1,033) was used for model development, and a testing dataset (n = 443) was used for internal validation. Another 167 patients with stroke recruited from October, to December, 2019, were enrolled in the dataset for external validation. A feature importance analysis was also performed to identify the significance of the selected input variables.ResultsFor predicting 30-day readmission after stroke, the ANN model had significantly (P &lt; 0.001) higher performance indices compared to the other models. According to the ANN model results, the best predictor of 30-day readmission was PAC followed by nasogastric tube insertion and stroke type (P &lt; 0.05). Using a machine learning ANN model to obtain an accurate estimate of 30-day readmission for stroke and to identify risk factors may improve the precision and efficacy of management for these patients.ConclusionUsing a machine-learning ANN model to obtain an accurate estimate of 30-day readmission for stroke and to identify risk factors may improve the precision and efficacy of management for these patients. For stroke patients who are candidates for PAC rehabilitation, these predictors have practical applications in educating patients in the expected course of recovery and health outcomes.